Regarding SPJ and iterate ranges, you may want to check out the paper he was presumably referring to here: http://research.microsoft.com/~s...excel/
index.htm I sent it to Giles a while back as a suggestion for a nifty new user story 
I see you are finally coming around to Haskell :D. It really is very nice. Has it's problems too of course but static typing is certainly not one of them. More on this later.
Regarding effects, it's of course useful to control them for building threaded code and software transactional memory systems.
However there is what is to my mind a more immediate benefit in that types are automatically inferred >documentation< about what functions may do. This means that e.g if you don't see an IO in the type you know it is a "sane" sort of function that doesn't futz around with arbitrary bits of the state you may depend on, making you much surer about composing and using functions in your code. This is a real help when working with 3rd party libraries and large code bases where you do not necessarily know everything about every function.
There are many more examples of the types as documentation view, which mainly comes about because the Haskell type system encodes so much more information (is "stronger" in some sense).
Right, I've gone on long enough
Hi Fuzzyman,
very interesting post indeed !
I'm afraid once you've seen and understood some Haskell code, you'll be jealous
of some of its nice features untill you swap to it in the near future.
I'm dreaming about having a Python parser module like Parsec for Haskell...
maybe this http://www.valuedlessons.com/200...ice-
syntax.html is just a start
See you.
I quited like the look of Haskell. We went through three hours of examples with Simon in his tutorial, so I get the basics. I don't think it will woo me permanently from Python but I hope it will make me a better programmer.
Great report Michael. Sorry I wasn't around for your presentation. I'm pleased to see Python represented at the ACCU conference, even now that the Python UK conference is a standalone event. One of the great things about ACCU is its breadth of scope, as I think is reflected in your report.
One small correction. I think it's "Simon Peyton Jones" (no hyphen). I mention this since I made the mistake in the first version of my own report, and the same mistake appears on the ACCU website.
Thanks for your ACCU conference report. I very much enjoyed your Python session, even though I'm more of a C++ guy! BTW, Sam Saariste was the one in the audience that pointed out a corner case that Andrei's identity function couldn't handle. http://www.voidspace.org.uk/
pyth...x.shtml#e958id5 For an overloaded function f and a function pointer fp, fp = identity(f) would not compile, even if fp = f would.
Commenting by HaloScan
